Common Installer Issues
The WireGuard module is not loaded while Secure Boot is enabled
Symptom: A "reboot required" warning at the end of the installation; modprobe wireguard fails.
Check:
Linux CLI | [root@server~]# mokutil --sb-state |
|---|
Solution: Disable Secure Boot from the firmware/UEFI settings of the server/VM. This cannot be changed with a plain reboot from inside the operating system; it is a firmware setting that requires a power cycle.
wg-quick fails to start with the "Key is not the correct length or format" error
Symptom: The service is in the "failed" state, and this error is shown in the journalctl/systemctl status output.
Solution: The peer public key field was left empty and the automatic fetch also failed, so an empty placeholder remained in the configuration file instead of a real key. As explained in Section 9, enter the real key on the /peer-key page, or have it fetched automatically with the IP address of the peer.
wg-quick gives the "Address already in use" error
Symptom: ip link add succeeds, but the IP assignment step fails.
Check:
Linux CLI | [root@server~]# ip link show type wireguard [root@server~]# wg show |
|---|
Solution: The WireGuard interface of another role tested earlier on the same server is still up and holds the same IP address (e.g., if the RPAM Connector and then the RPAM Portal were tested on the same VM). Stop the service of the unused interface: systemctl stop wg-quick@<name>.service
The Preflight Check shows packages as "missing" although the RPMs exist
Solution: The local repository metadata (repodata) is old/incomplete, or the package name is wrong (e.g., on RHEL 8 / Rocky Linux 8, "python36" should be used instead of "python3"). If new RPMs were added under files/*/rpms, the installation itself regenerates the repodata automatically; if the problem persists, collect the RPMs again with tools/collect-rpms.sh.
Multiple roles are tested on the same server and the results look inconsistent
Solution: The system works with the "one host = one role" logic. When a host is reconfigured for a different role, the services/interfaces of the previous role are not cleaned up automatically. For real/persistent tests, install each role on a separate server. If sequential tests are done on the same server, stop the WireGuard interface of the previous role manually before changing the role.
rpam-installer.service does not start (Method B)
Symptom: systemctl status rpam-installer shows that /pam/rpam-mtc/run.sh cannot be found.
Solution: The service file starts the wizard from the fixed path /pam/rpam-mtc/run.sh. Extract the package under /pam (see Section 4.2), or update the ExecStart line of /usr/lib/systemd/system/rpam-installer.service and run systemctl daemon-reload.
The peer public key cannot be fetched automatically
Symptom: "Could not fetch a public key from {IP}" error on the /peer-key page, or a "could not be auto-fetched" warning at the end of the installation.
Solution: The key is fetched from http://{real IP of the peer}:8443/api/public-key. Make sure that the RPAM MTC Installer is running on the peer, the peer has completed its own installation, and TCP 8443 of the peer is reachable. Otherwise, enter the key manually (see Section 9).
"Peer public key was not provided" warning during the RPAM Portal installation
Solution: This field is intentionally optional ("leave empty to auto-fetch or set later"). If the peer (RPAM Connector) is not installed yet or its real IP address was not entered, this warning is normal and is not an error. Apply Section 9 after the peer is installed.